Impact of real-time use of artificial intelligence in improving adenoma detection during colonoscopy: A systematic review and meta-analysis

Area of Science:

• Gastroenterology
• Medical Artificial Intelligence
• Endoscopy

Background:

• Deep neural networks (DNN) drive rapid advancements in artificial intelligence (AI).
• Recent randomized controlled trials (RCTs) explored AI integration in colonoscopy, focusing on adenoma detection rates (ADRs) and polyp detection rates (PDRs).

Purpose of the Study:

• To systematically review and meta-analyze RCTs comparing AI-assisted colonoscopy (AIAC) with conventional colonoscopy (CC).
• To determine if AIAC offers statistically significant improvements in detection rates, justifying its clinical adoption.

Main Methods:

• A comprehensive literature search identified RCTs comparing AIAC and CC.
• Primary outcome: Adenoma Detection Rate (ADR). Secondary outcomes: Polyp Detection Rate (PDR) and total withdrawal time (WT).
• Meta-analysis included six RCTs with 5058 participants undergoing average-risk screening colonoscopy.

Main Results:

• AIAC demonstrated significantly higher ADR compared to CC (33.7% vs. 22.9%; OR 1.76, 95% CI 1.55-2.00).
• AIAC also showed significantly higher PDR (45.6% vs. 30.6%; OR 1.90, 95% CI 1.68-2.15).
• Total withdrawal time was slightly higher with AIAC (MD 0.46 minutes).

Conclusions:

• AIAC utilization leads to increased adenoma and polyp detection.
• The findings support the clinical adoption of AIAC for enhanced colonoscopy screening.